Introduction
Cerebral amyloid angiopathy (CAA) is a vasculopathy characterized by the accumulation of amyloid-β (Aβ) in cerebral vessel walls. CAA is associated with an increased risk of cognitive decline and intracerebral hemorrhage (ICH) [1,2]. We have recently shown that approximately a quarter of the general elderly population has moderate-to-severe CAA pathology [3]. This underlines the importance of considering CAA in the differential diagnosis of patients presenting with cognitive decline or with transient neurological symptoms, which could indicate CAA-related transient focal neurological episodes (TFNEs). TFNEs are often mistakenly diagnosed as transient ischemic attacks, migraine aura, or focal seizures [4]. Interestingly, it has been suggested that superficial siderosis may induce seizure activity resulting in cortical spreading depression, which can cause focal seizures manifesting as TFNEs [5]. Avoiding misdiagnosis in patients with CAA is crucial, since the use of antithrombotic medication in CAA patients might be associated with an increased risk on ICH [6].
Alzheimer’s disease (AD) and CAA pathology frequently cooccur: moderate-to-severe CAA pathology is present in almost 50% of AD patients [3]. Recently, this has become increasingly important, as increased vascular Aβ deposition and subsequent local inflammation can occur as a frequent side-effect of anti-Aβ immunotherapy [7]. With the US Food and Drug Administration (FDA) approval of aducanumab [8] and lecanumab [9] as treatment for AD (The FDA approval of donanemab has been delayed to convene an advisory committee meeting to discuss the safety and efficacy data as of March 2024), proper awareness of the high prevalence of CAA has become even more important.
A definite diagnosis of CAA requires neuropathological postmortem investigation of brain tissue. Clinically, probable or possible CAA can be diagnosed using the Boston Criteria 2.0, which are based on the presence of strictly lobar hemorrhagic lesions (ICH, cerebral microbleeds, cortical superficial siderosis [cSS], or convexity subarachnoidal hemorrhage) and associated white matter characteristics (severe perivascular spaces in the semioval center or white matter hyperintensities in a multispot pattern), in combination with clinical symptoms of ICH, TFNEs, or cognitive impairment [10]. These criteria are most accurate in patients presenting with ICH (sensitivity 90%, specificity 93%), and have a lower sensitivity (55%) and similar specificity (96%) to diagnose CAA in patients with presentations other than ICH [10]. Insight into the prevalence of CAA in different ethnicities may be helpful to estimate the a priori chance of CAA in individual patients. This is especially relevant in the light of risk assessment before treatment of AD patients with immunotherapy, as AD patients with more severe CAA are at increased risk of developing side effects [11]. Furthermore, more insight into the etiology of ICH or cognitive impairment in ethnic groups may inform tailored prevention measures such as intensified cardiovascular risk factor management.
Few studies have investigated the differences in CAA prevalence in East-Asian versus Western populations. It has been suggested that the proportion of CAA-related ICH is lower in East-Asian populations than in Western populations [12], and that prevalence and severity of CAA pathology are lower in East-Asian populations [13,14]. A comparison of six East-Asian studies to four Western studies showed lower age-specific prevalence rates of CAA pathology in East-Asian versus Western populations; however, formal statistical assessment was not performed [15]. A more recent study on the prevalence of strictly lobar microbleeds in East-Asian versus Western populations did not show a difference in the prevalence and number of strictly lobar microbleeds, although a sensitivity analysis showed a trend towards higher prevalence of multiple strictly lobar microbleeds in Western populations [16]. This study also found a higher prevalence of strictly deep (a marker of deep perforator arteriopathy or arteriolosclerosis) and mixed microbleeds in East-Asian populations.
We set out to investigate the geographical differences of CAA prevalence in more detail, by performing a systematic review and meta-analysis to compare the prevalence of both CAA pathology and neuro-imaging markers associated with CAA in East-Asian versus Western populations.
Methods
For this systematic review and meta-analysis, we followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.
Search strategy and selection criteria
We updated our previous comprehensive search strategy (performed in March 2018 and updated in June 2019) in Embase and PubMed [3] on February 17, 2022. The search syntax included the keywords “cerebral amyloid angiopathy,” “cerebral hemorrhage,” “neuroimaging,” “neuropathology,” “amyloid-beta,” and “microbleeds.” Controlled search terms (Medical Subject Headings [MeSH] term) were combined with free text words. The reference lists of eligible studies and relevant reviews were searched for additional potentially relevant studies. We applied neither date nor language restrictions; papers were translated when necessary. References were imported into Endnote 20, which was used to remove duplicates. The protocol for this review, to which we now added the geographical subgroup analyses, was registered in the International Prospective Register of Systematic Reviews (PROSPERO; registration number CRD42018093159).
Inclusion and exclusion criteria
Primary research papers were eligible for inclusion if they described one of the following study populations: (1) general population (community-dwelling elderly, or consecutive autopsy series in case of a subset of pathology studies), (2) cognitively healthy elderly (no mild cognitive impairment or dementia, no stroke), (3) patients with AD (either clinically or pathologically diagnosed), (4) patients with ICH (irrespective of location), or (5) patients with lobar ICH. If a study reported on more than one of these study populations and segregation of data was not possible, the study was excluded. We included papers that reported summary estimates on at least one of the following outcome parameters: (1) CAA prevalence according to neuropathological assessment, (2) CAA prevalence according to the (modified) Boston criteria (v1.0 or v1.5) [17,18], (3) the prevalence of strictly lobar cerebral microbleeds, or (4) the prevalence of cortical superficial siderosis. Other inclusion criteria were: (1) study population comprised at least 10 subjects, (2) mean age (or median age, if mean age was not reported) of the population of ≥55 years, and (3) clearly defined diagnostic criteria to detect CAA which included the use of either neuropathology or MRI (T2* or susceptibility-weighted imaging [SWI]). Studies were excluded if they were (1) reviews, conference abstracts, commentaries, editorials, policy reports; (2) primarily focused on other pathologies as a cause of hemorrhagic neuroimaging markers, such as central nervous system malignancy, vascular malformation, excessive warfarin use, antecedent head trauma or ischemic stroke, vasculitis, blood dyscrasia or coagulopathy; or (3) focused on patients with isolated convexity subarachnoid hemorrhage. If multiple papers reported on overlapping parts of the same cohort, the study reporting on the largest population was included. Finally, we selected the papers that reported on East-Asian study populations (China, Japan, South Korea, and Taiwan), and those reporting on Western study populations (from Europe and North America).
Data extraction and analysis
Data extraction was performed in Covidence systematic review software (Melbourne, Australia) [19] by two independent authors as previously described [3]. Quality of the studies was assessed using an adapted and combined version of the quality assessment tools by Hoy et al. [20] and the Newcastle-Ottawa scale [21] as previously described [3]. A lower score corresponds to higher quality and the maximum possible score was 18 points. The median quality assessment scores with interquartile ranges (IQR) for studies on neuropathology and microbleeds were calculated separately.
For the CAA pathology studies, moderate-to-severe CAA was considered the primary outcome, but also data on mild-to-severe CAA (including all CAA grades) was extracted. When the Boston criteria were used for CAA diagnosis, probable CAA was considered the primary outcome. In addition to the information on the prevalence of strictly lobar microbleeds and cortical superficial siderosis, we also collected information on prevalence of deep and mixed microbleeds. Not all outcomes were available for every domain. Statistical analyses were performed using the “meta” (version 7.0-0) and “metafor” (version 4.4-0) packages of R (version 4.1.3; R Foundation for Statistical Computing, Vienna, Austria). Pooled prevalence estimates of CAA were calculated as previously described [3], using a DerSimonian-Laird random-effects model on Freeman-Tukey double arcsine-transformed data. In the same way, pooled prevalence estimates of hypertension were calculated. Heterogeneity was quantified using I2 statistics [22] and tested using Cochran’s Q. Potential geographical differences were assessed by univariable meta-regression analysis with geographical region as modifier and the prevalence in the two regions as outcome. In addition, a multivariable model with both geographical location and mean age (or median or midpoint of range, if mean was not reported) were included as modifiers. A P-value of less than 0.05 was considered statistically significant.
We have previously shown that CAA prevalence estimates are not influenced by the choice of random-effects model (as generalized linear mixed models did yield similar estimates), outliers (as assessed by influence analyses), low-quality studies (as sensitivity analysis including only high-quality studies yielded comparable estimates), and reporting bias (as assessed by inspection of funnel plots and Egger’s tests) [3]. Therefore sensitivity analyses were not performed for the geographical subgroups.
Results
The combined searches resulted in 12,257 unique records. After full-text screening, a total of 196 studies were included that fulfilled the inclusion criteria (Figure 1, Table 1, and Supplementary Tables 1-19). Of these studies, 76 reported on European and 68 on North-American study populations (in total 56,788 participants, and one study pooled a European with a North-American population) [23] and 53 reported on East-Asian study populations (24,920 participants). Ten studies were conducted in other countries (including one study of which we could not find out where the participants came from despite contacting the authors), these were not included into the analyses (Figure 2). The median quality assessment score of East-Asian studies was 2.5 (IQR 1.0-4.5) and of Western studies 4.0 (IQR 2.25-6.0) (Table 2). The median quality assessment score of imaging studies was 2.0 (IQR 1.0-3.625) and of pathology studies 4.0 (IQR 2.5-6.0) (Table 2). Prevalence rates of hypertension were comparable in East-Asian and Western studies (Table 3). Details regarding MRI slice thickness, field strength, and use of SWI can be found in Table 4.
In the general population, prevalences of mild-to-severe CAA (East-Asian: 28.5% vs. Western: 46.0%) and moderate-to-severe CAA (East-Asian: 22.4% vs. Western: 23.5%) (Supplementary Figure 1) were similar in both geographical locations (Table 1). The prevalence of strictly lobar microbleeds was significantly lower in the East-Asian (5.6%) compared to the Western (11.4%, P=0.020) (Supplementary Figure 2) general population. However, East-Asian cohorts were on average 6 years younger compared to Western cohorts, and after inclusion of age into the regression model the geographical difference disappeared (P=0.17). East-Asian cohorts had a higher prevalence of deep microbleeds (5.7%) compared to Western cohorts (2.7%, P=0.008 [P<0.0001 in the multivariable model]) as well as a tendency towards a higher prevalence of mixed microbleeds (3.4% vs. 1.7%, P=0.092 [P=0.002 in the multivariable model]).
In cognitively normal elderly, prevalence of mild-to-severe CAA (33.2% vs. 30.7%) did not differ between East-Asian and Western cohorts. No East-Asian studies were included that specifically reported on moderate-to-severe CAA (Table 1 and Supplementary Figure 3). There was a significantly lower prevalence of strictly lobar microbleeds (2.6% vs. 11.4%, P=0.001) (Supplementary Figure 4) in East-Asian versus Western cohorts of cognitively normal elderly that on average were of comparable age. However, this difference was no longer present when age was taken into account (P=0.056). The prevalence of mixed microbleeds and strictly deep microbleeds did not differ between East-Asian and Western studies. The prevalence of cSS was similar in East-Asian (1.0%) and Western (0.6%) cohorts of cognitively normal elderly.
In patients with AD, the prevalence of mild-to-severe CAA (83.3% v.s 77.5%) and moderate-to-severe CAA (55.4% vs. 44.1%) (Supplementary Figure 5) was similar in East-Asian and Western cohorts (Table 1). The prevalence of strictly lobar microbleeds (22.2% vs. 23.8%) (Supplementary Figure 6), mixed microbleeds (5.7% vs. 6.3%), and strictly deep microbleeds (5.9% vs. 5.6%), and cSS (3.9% vs. 4.3%) was similar as well. Including age into the regression model did not alter the results.
In patients with ICH, the prevalence of mild-to-severe CAA (27.0% vs. 27.1%) and moderate-to-severe CAA (11.8% vs. 26.3%) (Supplementary Figure 7) did not differ between East-Asian and Western cohorts (Table 1). The prevalence of strictly lobar microbleeds was lower in East-Asian (10.2%) compared to Western (24.6%, P<0.0001 [P=0.008 in the multivariable model]) (Supplementary Figure 8) cohorts of ICH patients, with East-Asian cohorts being on average 7 years younger. In contrast, the prevalence of mixed microbleeds was higher in East-Asian cohorts (40.6%) compared to Western cohorts (20.6%, P=0.045), but not in the multivariable model (P=0.15). There was a tendency towards a lower prevalence of probable CAA according to the Boston criteria in East-Asian versus Western countries (9.5% vs. 27.4%, P=0.026), albeit not in the multivariable model (P=0.14). The prevalence of strictly deep microbleeds (21.8% vs. 16.7%, P=0.42) and of cSS (10.1% vs. 16.8%, P=0.13) was similar.
In patients with lobar ICH, the prevalence of mild-to-severe CAA (52.2% vs. 51.9%) and moderate-to-severe CAA (49.7% vs. 59.0%) did not differ between East-Asian and Western cohorts (Table 1). No East-Asian imaging studies were included, and therefore, no comparison between the prevalence of imaging markers could be made.
Discussion
We demonstrate that the prevalence of CAA pathology does not differ between East-Asian and Western cohorts reflecting the general population, cognitively normal elderly, patients with AD, patients with ICH, and patients with lobar ICH. Furthermore, we found that in the East-Asian general population, cognitively normal elderly, and patients with ICH, the estimated prevalence of strictly lobar microbleeds was lower compared to Western cohorts, although this may be partly explained by a lower mean age in East-Asian cohorts.
Only a few studies to date have compared the prevalence of CAA pathology in East-Asian and Western countries [13-15]. We report comparable prevalence estimates of moderate-to-severe CAA in the East-Asian (22.4%) and Western (23.5%) general population. In addition, potential geographical differences regarding the prevalence of CAA imaging markers have received only limited attention. In an individual participant meta-analysis corrected for age, deep/infratentorial and mixed microbleeds were more commonly present in East-Asian (e.g., from Japan, South Korea, and China) versus Western (e.g., from Iceland, Australia, and the USA) stroke-free individuals aged 55-75 years (odds ratio [OR] 2.78, 95% confidence interval [CI] 1.77-4.35, P<0.002) [16]. In contrast, the prevalence of strictly lobar microbleeds did not differ (OR 0.70, 95% CI 0.29-1.72, P=0.44) between East-Asian and Western individuals, although a trend was observed in a sensitivity analysis assessing only the prevalence of multiple strictly lobar microbleeds (OR 0.43, 95% CI 0.17-1.04, P=0.062). This individual participant data meta-analysis supports our study-level multivariable analysis in which East-Asian location was associated with a higher prevalence of mixed and strictly deep microbleeds in cohorts reflecting the general population. In our univariable, but not multivariable model, Western geographic location was associated with a higher prevalence of strictly lobar microbleeds in the general population, indicating at best a weak association between geographic location and the occurrence of strictly lobar microbleeds. Most likely, the lower prevalence of strictly lobar microbleeds in East-Asian study populations compared to Western study populations is due to a higher incidence of mixed microbleeds.
We show that the prevalence of CAA pathology and strictly lobar microbleeds does not differ between East-Asian and Western cohorts of patients with AD. This finding is relevant given recent developments in the field of AD treatments. In 2021, aducanumab was approved by the FDA as a treatment for AD, and in January 2023, the FDA approved lecanemab [8,9]. In addition, lecanemab (but not aducanumab) was approved by the Ministry of Health, Labour and Welfare in Japan in September 2023 [24], and by the National Medical Products Administration in China in January 2024 [25]. However, the safety and efficacy of aducanumab and lecanemab remain controversial [26-28]. Both immunotherapies come with frequent adverse effects in the form of amyloid-related imaging abnormalities (ARIA). ARIA is thought to reflect local inflammation associated with vascular deposition of Aβ that has been released as a result of antibody-mediated breakdown of neuritic plaques [2]. This leads to vasogenic edema (ARIA-E) and/or microbleeds, cortical superficial siderosis, and ICHs (ARIA-H). ARIA is asymptomatic in about 75% of patients, but may lead to headache, confusion, nausea, visual disturbances, and dizziness [27,29]. Immunotherapy may exacerbate pre-existing CAA, which is present in many patients with AD. Therefore, patients with AD and concomitant moderate-to-severe CAA are at higher risk of developing ARIA. It is therefore recommended to exert extreme caution when prescribing immunotherapy treatment in patients with AD with concomitant moderate-to-severe CAA, especially when they have other risk factors for ICH, such as anticoagulant use [11]. Furthermore, the apolipoprotein E (APOE) ε4 allele is a risk factor for severe CAA, since it has been found that APOE ε4 carriers have more severe CAA, even when controlling for the extent of AD pathology [30]. In addition, it has been found that ARIA-E incidence is APOE ε4-dependent [11]. The FDA label lecanemab contains the recommendation to test for APOE ε4 status prior to initiation of treatment, and discuss the accompanied risk of ARIA with patients. It remains a challenge to clinically establish the severity of concomitant CAA, but it is important to note that in the phase III lecanemab trial, AD patients with four or more microbleeds, cortical superficial siderosis, and/or an ICH of >1 cm were excluded [9]. In addition, it was recently found that the presence of two to four microbleeds more than doubled the risk of ARIA-E in the phase II and III donanemab trials [31]. Our data indicates that CAA is equally prevalent in East-Asian compared to Western patients with AD. Therefore, screening for CAA and caution is warranted when prescribing immunotherapy to East-Asian as well as to Western patients with AD.
It has been suggested that the proportion of ICH caused by CAA is lower in Asian compared to Western countries [15]. This potential difference has been systematically assessed by studying consecutive patients with spontaneous ICH at two stroke centers during the same time period; one in the UK (279 patients) and one in Japan (214 patients) [12]. Patients from the Japanese center had lower odds of CAA-related ICH (OR 0.55, 95% CI 0.31-0.98) [12] according to the Edinburgh criteria [32]. The authors observed proportions of CAA-related ICH of 10.2% in patients of East-Asian ethnicity and 23.8% in patients of white ethnicity. As the incidence of ICH is twice as high in East-Asian compared to white populations (51.8 vs. 24.2 per 100,000 person-years) [33], the authors estimated that the incidence for CAA-related ICH is comparable for East-Asian and white populations (5.3 vs. 5.8 per 100,000 person-years). In contrast, the incidence of other types of ICH (mainly associated with deep perforator vasculopathy) was 2.5-fold higher in those of East-Asian ethnicity compared to those of white ethnicity (46.5 vs. 18.4 per 100,000 person-years) [12]. This indicates that the lower proportion of CAA-related ICH in East-Asian individuals with ICH is due to a higher incidence of ICH related to deep perforator vasculopathy rather than to a lower incidence of CAA-related ICH. This may also contribute to our finding that the prevalence of strictly lobar microbleeds was 2.5 times lower in East-Asian cohorts of ICH patients than in Western cohorts. Unfortunately, the East-Asian studies reporting on MRI markers of CAA that we included in our meta-analysis provided no details on ICH location.
Hypertension is considered a risk factor for deep microbleeds and deep ICH [34,35]. However, in previous studies demonstrating a higher prevalence of deep microbleeds [16] in East-Asian (compared to Western) stroke-free individuals and a higher proportion of deep ICH in East-Asian (compared to white) ICH patients [12], the prevalence of hypertension did not differ between East-Asian participants and their Western counterparts. Similarly, we did not find evidence for a higher prevalence of hypertension in East-Asian cohorts, whereas deep and mixed microbleeds were more prevalent in the East-Asian general population and patients with ICH. It is possible that the increased prevalence of deep and mixed microbleeds might also be due to an increased susceptibility to develop hemorrhagic brain lesions in East-Asian individuals [12,36]. Whereas the exact underlying mechanisms of this increased susceptibility are yet unknown, East-Asians carrying an APOE ε2 or APOE ε4 allele have increased susceptibility to develop hemorrhagic lesions compared to Europeans with similar APOE polymorphisms [37,38]. Another explanation may be that epidermal growth factor-like repeat (EGFr) cysteine-altering NOTCH3 mutations are more common in East-Asian populations than in Europe (9/1,000 vs. 3/1,000) [39]. Such mutations may result in a mild CADASIL (cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy)-phenotype clinically indistinguishable from sporadic small vessel disease. In addition, a meta-analysis found that the occurrence rate of ICH in CADASIL is higher in CADASIL patients from Asia (17.7%) compared to CADASIL patients in Europe (2%) [40].
Strengths of this study include our comprehensive search and selection strategy, resulting in the largest dataset to date on available papers that report the prevalence of CAA. Additionally, we included papers both on CAA pathology as well as on radiological markers for CAA, enabling us to determine a reliable estimate of the prevalence of CAA in Western and East-Asian countries. Limitations of this study include the heterogeneity in methods that have been used to assess CAA pathology [3], as well as potential differences between MRI acquisition protocols, of which we previously demonstrated this influenced the detection of microbleeds [3]. Also, included studies varied in quality, which may have introduced a bias. Another limitation is the large age differences (up to 15 years) between East-Asian and Western cohorts reflecting the general population and ICH patients. Microbleed prevalence has been reported to be associated with age [3,36,41]. We corrected for age in an additional statistical model but due to the small number of studies, and because we did not have individual patient data, no firm conclusions can be drawn from this model. However, the observation that the East-Asian populations were on average younger and that the prevalence of CAA pathology and of strictly lobar microbleeds were lower in East-Asians, may imply that the prevalence of these outcomes would be more similar to Western-based cohort when they would have been of similar age. Furthermore, our comprehensive search was designed to include all studies on strictly lobar microbleeds and we might not have included all available studies on deep or mixed microbleeds. Finally, we included studies based on the country they were conducted rather than ethnicity of the study populations. Therefore there may have been East-Asian participants in the Western cohorts and vice versa.
Conclusions
In this comprehensive meta-analysis, we show that the prevalence of pathologically established CAA is similar in East-Asian and Western countries, and radiological markers associated with CAA, i.e., strictly lobar microbleeds, are less prevalent in East-Asian cohorts of population-based individuals and cognitively normal elderly. The latter observation may be due to younger East-Asian cohorts, since after correction for age the difference was non-significant. In contrast, we show that the prevalence of strictly deep microbleeds and mixed microbleeds is higher in East-Asian population-based cohorts, even after correction for age. This indicates that deep perforating vasculopathy is more widely present in East-Asian populations and suggests that preventive measures are urgently warranted in these regions. Furthermore, caution should be employed when including East-Asian as well as Western patients with AD in immunotherapy trials considering similar prevalence estimates of pathological established CAA.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Supplement1
Print
